Google 发布 Gemma 4 QAT 检查点，可在消费级 GPU 和移动设备上本地运行且质量损失很小。其中 GGUF（Q4_0）检查点覆盖各尺寸与 drafter 模型以追求本地性能，定制移动方案通过混合精度把 Gemma 4 压缩到 1GB 以内，包含针对性 2-bit 解码层、优化 KV cache 和静态激活。
文章列出 Gemma 4 QAT 的 GGUF 与移动端两种量化方案，可用于判断本地部署的体积与质量取舍。
New @GoogleGemma 4 QAT (Quantization-Aware Training) checkpoints are here, so you can run models locally on consumer GPUs and mobile devices with minimal quality loss.
What’s new:
🔹 GGUF (Q4_0): Checkpoints: Max local performance across all sizes and drafter models
🔹 Custom Mobile Schema: We shrunk Gemma 4 down to less than 1GB for mobile devices by using a custom mixed precision schema designed for edge hardware (featuring targeted 2-bit decoding layers, optimized KV caches, and static activations)
By simulating compression during training rather than after (Post-Training Quantization), we've drastically reduced the memory footprint and accelerated decode speeds while preserving reasoning quality. blog.google/innovation-and-a…
来源：@googleaidevs · x.com